Data and code for the paper by Battistin and Sianesi:

"Misclassified Treatment Status and Treatment Effects: An Application to
Returns to Education in the UK"


1) DATA	(all data is in Stata format)

1. NCDS_Battistin_Sianesi.dta
All the variables have appropriate labels.
It just needs to be clarified that:
	qmab_* denote quintiles of Math tests at 7
	qmab2_* of Math tests at 11
	qvab_* of Reading tests at 7
	qvab2_* of Reading tests at 11.
d# denotes the #-th quintile, with #=5 being the highest.


2. lamdagrid.dta
The grid with the values for lambda for the binary treatment case.

3. mtlamdagrid5_90.dta
The grid with the values for lambda00 and lambda11 for the multiple treatment case,
with lambda21 = 0.05 and lambda22 = 0.90

4. mtlamdagrid1_95.dta
The grid with the values for lambda00 and lambda11 for the multiple treatment case,
with lambda21 = 0.01 and lambda22 = 0.95


2) CODE	(all in Stata)

There are two master do-files:

1. mebounds.do	-> code for deriving the raw estimates, as well as
				   the bounds and their CI's for BINARY treatments
				   
	calls: 	globalX.do
			meraw.ado
			mebounds.ado
			CIbounds.do
			hmboot.ado

			
2. mtbounds.do -> code for deriving the raw estimates, as well as 
				  the bounds and their CI's for MULTIPLE treatments
				  
	calls: 	globalX.do
			mtraw.ado
			mtbounds.ado
			CIboundsMT.do
			hmboot.ado



			